Inter Partes Review Statistics

By Jason Rantanen

This post summarizes data on inter partes review proceedings and appeals from the Patent Office.  Although the office publishes a monthly Patent Trial and Appeal Board Statistics packet, the narratives contained within that packet can create confusion as discussed in Michael Sander’s guest post earlier this year.  Below are some of the charts that I’ve developed based on the publicly available information to attempt to get a better handle on what’s going on in terms of case flow and outcome.

As you look at these charts, keep in mind the temporal issues associated with IPR proceedings.  Once a petition is filed, it typically takes about six months before an institution decision is made.*  If a petition is instituted, it will be about a year (and sometimes more) before the trial decision.  Thus, very few petitions filed in the last four months or so will have reached the point of institution decisions, and nearly all trial decisions issued in the last few months relate to petitions filed over a year (and probably around 18 months) ago.  Appeals of trial outcomes will be filed within 2-3 months depending on whether a rehearing is requested.

The consequence of this is that each successive piece of information relates to an earlier time period.  The institution rates shown in the first graph relate to the petitions filed about 5-7 months earlier; the trial outcomes shown in the second graph relate to petitions filed about 18 months earlier (and instituted about 12 months earlier), and the appeal numbers relate to trial outcomes from about 2-3 months earlier (with the additional caveat that these are all appeals from the Patent Office, not just appeals from IPRs).  These sequences are approximations, but they’re important in understanding the relationship between filed petitions, institution rates, trial outcomes and appeals.

Finally, note that at each stage, some percentage of the cases continuing on after the earlier stage will drop out due to settlement or for other reasons (such as joinders at the institution stage).  Thinking of the data as an inverted pyramid might be helpful.

Petitions Filed and Institution DecisionsIPR Petitions1

Petition filing and institution rate data indicates that both the rate of filing and rate of institution have plateaued.  Since June of 2014, the number of new petitions filed each month has ranged between 96 and 184, with most quarters falling between about 400 and 450 new filings.  The fourth quarter of 2015 and the first quarter of 2016 did see a drop, though, so perhaps we’ve seen the peak of IPR petition filings.

Trial OutcomesIPR Trial Outcomes

The most interesting thing to note about the above graph of trial outcomes, which I created based on the differences reported in the monthly statistical reports issued by the PTO, might be that recent months have seen a fall in the percentage of trials in which all claims were held unpatentable.  In March 2016, for example, while 40 trials resulted in all claims held unpatentable, twenty-one resulted in no claims held unpatentable.  And in May and June, only a bit over 50% of the trials resulted in all claims held unpatentable.


The below chart depicts the numbers of appeals docketed at the Federal Circuit arising from the PTO, the District Courts, and the International Trade Commission.  It illustrates a dramatic rise in the number of appeals from the PTO, with the number projected to surpass the number of appeals arising from the district courts this year.  It’s use in assessing IPR proceedings is limited due to the fact that the data does not differentiate between appeals from IPR proceedings and other appeals from the PTO.    I’d be interested in hearing from anyone who’s done the detailed breakdown of these appeals to know whether the standard hypothesis–that this rise is mainly due to appeals from IPR proceedings–is correct.

Appeals Docketed

*These numbers are based on the PTO’s own timeline, see 77 Fed. Reg. 48757 (Aug. 14, 2012), and data from LexMachina indicating the median time to institution decision is currently 185 days, and the median time to final decision is currently 540 days.

32 thoughts on “Inter Partes Review Statistics

  1. 6

    Looking at the last graph and the projected final data points, I cannot help but note that the Congressionally stated purpose of the IPR does not seem to have been met. In fact, it appears to have failed miserably.

    Granted, the numbers here are at the appeals level, and only serve as a rough proxy for original cases at the district court level, but consider:

    The projected drop or “savings” of cases in the last year is about 100.
    The projected drop or “savings” of cases over the last two years is about 30.
    The projected drop or “savings” of cases over the last three years is a negative amount.

    So, depending on how big of a slice of time, there appears to be NO (or extremely little) drop in workload – again, using the appeal as a proxy.

    But that’s only half the story.

    As noted, the actual TOTAL workload is drastically increasing.

    The three year snapshot is about 500 cases.

    Given the rationale for IPR as an alternative to the litigation path, and given the other Constitutional infirmities (yes, some may attempt to poo poo these very real concerns), it appears that these particular Means to whatever desired Ends may not be all that they are cracked up to be…

    Leastwise, for what they were sold as.

  2. 5

    link to

    A map display program system claim called for a “symbol generator.” The Federal Circuit affirmed a district court holding that the claim term was governed by 35 USC § 112(6) because the term “symbol generator” did not invoke known software modules, but was a coined term. As such, it was effectively a means-plus-function clause and therefore must have corresponding structure described the specification or was invalid as indefinite. Here because software was involved, the specification had to disclose an algorithm for how the symbols were generated. No such algorithm was described in the specification. All the specification did was say generate the symbols from a database (using a computer) without more. Thus it covered any and all software that might generate a symbol from a database – the very essence of functional claiming.

    The specifications of the patents-in-suit do not disclose
    an operative algorithm for the claim elements reciting
    “symbol generator.” The function of generating
    symbols must be performed by some component of the
    patents-in-suit; however, the patents-in-suit do not describe
    this component. Although the specification of the
    ’728 patent suggests that these symbols are generated via
    “a map database and a database of geographically referenced
    fixed locations . . . with a specified latitude and
    longitude[,] . . . [and] [a] database with the constantly
    updated GPS location,”’ 728 patent col. 3 ll. 35–41, this
    only addresses the medium through which the symbols
    are generated. A patentee cannot claim a means for
    performing a specific function and subsequently disclose a
    “general purpose computer as the structure designed to
    perform that function” because this “amounts to pure
    functional claiming.” Aristocrat Techs., 521 F.3d at 1333.
    Accordingly, because the specifications of the patents-in suit
    do not disclose sufficient structure for the “symbol
    generator” function and the asserted claims include this
    term, the asserted claims are indefinite under 35 U.S.C. § 112, ¶ 2.

  3. 4

    For evidence that the IPR framework should be expanded to include eligibility challenges, see today’s silly time-wasting result in RESPIRONICS, INC. v. ZOLL MEDICAL CORPORATION (Fed Cir 2016) where the prior art was distinguished solely on the content of data that is transmitted and stored by a system (in this instance, “patient compliance data”).

    Absurd and pointless.

    1. 4.1

      The actual decision of the Fed. Cir. here is that this IPR PTAB decision read the claim language too narrowly, and did so to avoid the prior art. [Note ironically that is the opposite of the usual complaint about IPRs using BRI!]:
      We hold that the Board erred in finding that the Owen
      reference does not disclose “patient compliance data.”
      Based on this error, we vacate the Board’s decision rejecting
      Respironics’ anticipation arguments as to claims 2, 4,
      5, 8, 9, 16, 19, and 20. We remand this case to the Board
      so that it may consider whether Owen satisfies these
      claims’ remaining limitations.”

      1. 4.1.1

        Right. The Board (presumeably at the behest of the patentee) made some logical leaps about what “patient compliance data” would comprise, content-wise.

        But the bigger and more important issue is that if the art taught a device that sends information to a remote database, one should never be able to get a patent over that device by reciting only (1) the old device and (2) some “new” information that is transmitted by the device. That’s exactly the kind of b0tt0m-feeding junk that 101 is intended to eliminate. And there’s plenty of that crxp that should be IPR’d out of existence in two seconds with the ultimate goal being that such patents are never asserted and (ideally) never granted.


          MM, haven’t read the case yet, but I agree with you that if the only thing a different from the prior art is the kind of data transmitted, then the patent effectively is on the data. People practicing the prior art infringe if the data transmitted is a particular kind of data.

          This would be like a patent on the player piano, old in the art, performing the new song. The infringement is playing the new song.


                Placing aside for the moment that software is not “data comprising numbers,”…

                Shall we revisit the simple set theory on the exceptions to the judicial doctrine of printed matter?

                Or do you need a primer in the difference between the Fine Arts and the Useful Arts?

                (hint: it’s related to the utility aspect of 101).

                1. Placing aside for the moment that software is not “data comprising numbers,”…

                  This isn’t a software claim. It’s a claim to a defibrillator device further comprising sensors and transmitter for transmitting data to (wait for it!) a database.

                  Shall we revisit the simple set theory on the exceptions to the judicial doctrine of printed matter?

                  Does your silly “set theory” allow people to claim a new “telephone system” based on the recitation of the “new” content of a conversation relayed over the system?

                  If so your silly “set theory” is even sillier than I thought.

                2. Your post is nothing but dust-kicking.

                  I have repeatedly asked you to join in on an actual discussion of my simple set theory explication of the exceptions to the judicial doctrine of printed matter, and each time you jeer as you run away from any actual discussion on the merits.

                  I posted hyperlinks and you wanted an actual reposting.
                  I gave youthe actual reposting, and you went silent.

                  So here now, you intrude on my point to Ned and engage in personal sniping of empty name-calling and attempting to take the point I make completely out of context.

                  Even sillier than you thought? What does that even mean? Since “what you thought” – in any substantive manner is not known as you have steadfastly refused to engage in any discussion on the merits, ALL that we have from you is the same empty ad hominem that is your trademark (and that has been documented on this blog AS your trademark for a full decade now).

                  Maybe instead of the empty name-calling and dissembling, you attempt to discuss things on the merit so that we really do know exactly why you feel something is “silly” as opposed to just hearing your feelings yet again.

    2. 4.2

      Actually, I am pretty sure that an Alice challenge should be allowed since the holding of Alice is that it was unconstitutional to grant the claims.

      So, that probably means (I don’t have the administrative case) that Alice challenges can be brought since I am pretty sure a Constitutionality can always be brought.

      Anyway, I think this could go to the SCOTUS. The only problem is that the SCOTUS denies that the holding of Alice is that it was unconstitutional to grant the claims, but anyone with a law degree can see that that is the holding.

    3. 4.3

      They are allowed in the post-grant proceedings.

      But, actually I agree that IPRs should be expanded to everything that can be challenged in a fed. dist. ct.

      1. 4.4.1

        Do you really think that you fit “in between” given your on the record statemtent that ALL software inventions are de facto not eligible…?


          Do you really think that you fit “in between”

          I don’t know or care. I could have written “because there’s only two choices!”. The important point, of course — and it’s one that “newhart” and your other fellow travelers should keep in mind — is that just because someone is opposed to patenting logic “on a computer” doesn’t mean that they “don’t give two hoots about computer-related inventions”, nor does it mean that everything “computer-related” is obvious.

          That seems like an incredibly obvious point and it’s kinda sad that anybody needs to make it. But then again, the luvvers of software patents have never been widely admired for their thoughtfulness and intelligence.



            That’s great an as you know, software is not logic.

            Check out the protection for a different aspect of software with copyright, and check yourself that you cannot obtain a copyright on logic.

            Then again, you already knew that.


            Prof. Rantanen,

            Serious question for you:

            How is ANY of what Malcolm posted at even remotely fitting to your “rules”…?

            Malcolm states “kinda sad” – and it is “kinda sad,” just not how he would spin it.

            “Don’t know or care” that he himself has painted himself with a view that is accurately reflected by newhart, and that he just will continue to post how he has posted for a decade now.

            Your “rule” is as your “rule” is enforced.

            Maybe you too are just [shrugging]…

    4. 4.5

      MM, to the extent that a type of data transmitted is the only distinction over the prior art, under the printed matter doctrine, such data, which is not eligible subject matter under 101, cannot serve to distinguish over the prior art unless it is shown that the data is functional with respect to the underlying subject matter, i.e., it must be used to functionally modify a process or system that would otherwise constitute patentable subject matter.

      From my reading the case, the data here, “patient compliance data,” is not used, but simply transmitted. It therefore cannot legally distinguish the claim over the prior art under the printed matter doctrine.

    5. 4.6

      Newhart, check out post 5 regarding software patents. That case will give you a headache.

  4. 3

    The only 66% current IPR institution rate indicated here means 34% initial wins for patent owners. Add in settlements as well as all the patent owners shown in the these charts as now getting favorable IPR decisions [for all or at least one infringed claim]. These actual numbers in which the patent owners claims survive IPRs are a far cry from the still constantly repeated false “death squad” statistics.
    Re Max’s question, sure, a losing IPR petition might affect some license or settlement negotiations values between reasonable parties. But since something like 80% of IPRs are by parties already being sued, and since only a very small percentage of patent suits are ended by the claims in suit being held invalid over prior patents or publications [vastly less than non-infringement or even 101 decisions], filing and losing an IPR cannot have much effect on average patent suit outcomes. Whereas filing and winning an IPR is highly effective in ending many patent suits, including all other and future suits against customers or others on the same [now eliminated] claims.

    1. 3.1

      Thanks Paul. Striking (for me) are your observations that:

      1) “…a very small percentage of patent suits are ended by the claims in suit being held invalid over prior patents or publications..”

      2) “…an IPR is highly effective in ending many patent suits, including all other and future suits against customers or others on the same [now eliminated] claims.”

      Small wonder then, the hot feelings IPR can arouse in some people.

      1. 3.1.1

        The hot feelings on this blog about IPRs have a lot more to do with personal lost patent litigation billing opportunities.


          Or not.

          That Constitutional issue has ZERO to do with lost litigation billing opportunity.

          Just saying.

    2. 3.2

      Look, Paul, since only 33% of validity trials relying only on references result in a judgment of invalidity, the double invalidity rate in the PTO remains telling.

  5. 2

    Intuitively, one might surmise that many of the earliest cases concern “low-hanging fruit” vulnerable to assaults on validity but, as time goes by, the number of such cases dwindles. Further, as the library of decisions rejecting validity attacks grows, so does the realisation that to fail in an assault on validity can have consequences worse than not making the assault in the first place.

    See what I’m suggesting, that the numbers are heading towards a reasonable steady state, just like with oppositions at the EPO, since 1978. This is not rocket surgery, is it?

    1. 2.1

      Conceptually, I “get” the “keep the powder dry” thought that you share – but is it really more than just a concept? I tend to doubt that anyone has done any analysis to support or reject the concept, and at this point, it is just not possible to do more than just offer it as a supposition, still needing something more to establish its veracity.

      1. 2.1.1

        But, aside from that, and apart from the log-jam at appeal level, the sky is not actually falling, is it?


          The sky may not be falling – but that does not mean that the ship might not still be sinking.

          (that would be the good ship USS AIA Titanic)

  6. 1

    I read in a commentary on the Indian justice system where the average judge has thousands of cases on his docket and where the phrase “justice delayed is justice denied” is in common currency among the population, that the average judge should have no more than 300 cases on his docket that any one time. Now considering that until recently the combined total of patent cases from all sources was around 600 or less, this meant that the average Federal Circuit judge would have about 150 patent cases a year before him or her. Again assuming that the patent cases represent about half the Federal Circuit docket, the rest of it being appeals from other agencies involving nonpatent issues, this would line up with the up estimate that the average Federal Circuit judge had 300 cases a year before him or her.

    Dividing 300 by 50, assuming at least a two week vacation, we have six cases per week to have to be decided by each judge, but only two cases per week be written by a judge (panels of three). That is still some work load, which helps explain why so many decisions are simply rendered under rule 36.

    But as we see, the appeals from the patent office are growing so that the number of appeals considered by each judge is also growing. I would think that they are increasingly likely to use rule 36 simply to stay afloat. I also think this helps explain why the Federal Circuit ruled as it did in Cuozzo and Achates that it did not want to hear appeals regarding institution decisions at all. It simply did not have time to consider them.

    1. 1.1

      Ned, no question that IPRs have increased the current Fed. Cir. case load.
      But note that the number of issues and the size of the record below are normally both vastly smaller in an IPR appeal than in a D.C. trial appeal, with the exception of D.C. preliminary 101 case disposals.
      Also it is still too soon to tell when, and at what level, a “steady state” number of D.C. suits and appeals therefrom will be reached. Not until after the full impact of the Sup. Ct. on pending and future D.C. cases from Alice & Mayo, the increased attorney fee sanction risks, the increased enhanced damages and ambiguity decision risks. In addition to all the pending IPRs, CBMs, PGRs, and supplemental reexaminations.

Comments are closed.